If you use ChatGPT on the free plan — for recipes, for emails, for that last-minute question before a meeting — you've probably hit the wall. "You've reached your daily limit." You close the tab. You wait. Or you just don't get the answer.

That wall is going away. On August 6, OpenAI announced that free users are getting unlimited text conversations, rolling out starting the week of August 10. And while doing it, they swapped the model underneath the free product — out with GPT-5.5 Instant, in with GPT-5.6 Luna, a noticeably more capable model. Same product, better engine, no cap.

The thing I want to talk about is a button nobody is paying attention to.

What actually changed

Here's the short version.

You open free ChatGPT today and get GPT-5.6 Luna instead of the older GPT-5.5 Instant. Luna is the mid-tier model in OpenAI's current 5.6 family — the same one that was being sold via API at $0.20 per million tokens after a price cut last week. It's measurably better at following complex instructions, writing in different styles, and not making things up. An actual upgrade, not just a version number bump.

Starting the week of August 10, the daily message cap disappears for text. You can ask as many questions as you want. If you have a three-hour research session, a long drafting project, an afternoon of questions to work through — you can just do it. OpenAI says 1 billion people use ChatGPT every week, most of them on the free plan. So this is a real change for a lot of people.

The catch: image generation, file uploads, and the various other tools — web search, memory features — still have limits. Removing the text cap doesn't unlock everything. It just removes the most frustrating constraint for everyday use.

The Think button

Free users are also getting a Think button. You tap it on a specific message before sending, and GPT-5.6 Luna pauses to reason through the problem before answering — instead of firing the first response it generates.

This is like the difference between asking a question in passing versus asking someone to sit down and think about it before replying. Same chatbot. Different mode. The Think button puts the "sit down and think" mode in free ChatGPT for the first time.

Until now, extended reasoning was part of what you paid for. Tapping Think on a hard question — a medical question, a legal question, a complex work decision — was a Plus feature. It isn't anymore. And it's per-message, not a global setting, so you can leave it off for quick things and switch it on when you need the model to work harder.

Source spread

• OpenAI — Improving GPT-5.6 Sol in ChatGPT and expanding Luna access — hype. The primary announcement. Confirms the 1 billion weekly users stat, the Luna upgrade, unlimited text, and the Think button timeline. Also updates GPT-5.6 Sol for Plus/Pro.
• TechCrunch — Unlimited text chats for free users — builder. Solid on the rollout specifics; flags that limits on non-text tools remain.
• Developers Digest — Sol retune and Luna default — skeptic. Points out that Luna is the smaller model in the 5.6 family; free users gained in quantity but didn't gain access to the frontier.

What's actually better

• The daily cap was the most common complaint from free-tier users. Removing it, even if other tool limits stay, is a real quality-of-life change.
• GPT-5.6 Luna is a meaningful step up from GPT-5.5 Instant on writing quality, nuance, and instruction-following.
• The Think button changes who has access to extended reasoning. That used to be a paid feature. Now it's free, per message.
• OpenAI is reacting to competition. Claude has a generous free tier. Gemini has a generous free tier. This is what competitive pressure looks like in consumer AI: the floor gets raised.

What still deserves scrutiny

• Luna is the smaller model in OpenAI's current lineup. GPT-5.6 Sol — what Plus and Pro subscribers use — is noticeably more capable. "Unlimited free ChatGPT" still isn't the same as the best model.
• Image generation, file uploads, and tool access are still rationed. Those aren't minor features for many tasks.
• "Unlimited text chats subject to abuse guardrails" is a qualifier doing unknown work. The announcement doesn't define what triggers a limit under guardrails.
• This removes a reason to upgrade for casual users. Whether OpenAI planned for that or is betting on upselling to more capable tiers is worth watching.

What to do about it

Three adjustments worth making this week.

• Try the Think button on something that actually matters. Not "what should I make for dinner." The next time you ask ChatGPT a hard question — medical, financial, something with real stakes — tap Think before sending. It's a different experience from the default mode.
• Check whether your limit has lifted. The rollout is happening over the week of August 10. If you still hit a cap, wait a few days; the upgrade may not have reached your account yet.
• Re-evaluate whether you need the paid tier. If the main reason you were paying $20/month was to avoid the daily text cap, re-read what free now includes. For casual use, you may not need to pay anymore. For image generation, file uploads, or constant heavy use, the paid tier still earns its cost.

Further reading

• OpenAI — Improving GPT-5.6 Sol in ChatGPT and expanding access to GPT-5.6 Luna — primary announcement
• OpenAI — Advancing price-performance with GPT-5.6 — the July 30 price cut that made Luna viable as a free-tier model
• TechCrunch — Unlimited text chats for free users — rollout specifics